Skip to content

Qwen2.5-Omni: Update modeling_qwen2_5_omni.py to fix error when loading quantized weights with AutoAWQ. - #38013

Merged
ArthurZucker merged 2 commits into
huggingface:mainfrom
tiger-of-shawn:main
May 20, 2025
Merged

ArthurZucker merged 2 commits into
huggingface:mainfrom
tiger-of-shawn:main

Conversation

@tiger-of-shawn

Copy link
Copy Markdown
Contributor

Qwen2.5-Omni: Update modeling_qwen2_5_omni.py to fix loading quantized weights with AutoAWQ.

What does this PR do?
while loading AWQ quantized Qwen-Omni model with AutoAWQ, there will be an error as below:

Traceback (most recent call last):
File "/nas/yuehu/NEW/AutoAWQ/./examples/quantize_qwen_omni.py", line 188, in
inference_quantized_model(quant_path)
File "/nas/yuehu/NEW/AutoAWQ/./examples/quantize_qwen_omni.py", line 123, in inference_quantized_model
model = AutoAWQForCausalLM.from_quantized(quant_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/nas/yuehu/NEW/AutoAWQ/awq/models/auto.py", line 125, in from_quantized
return AWQ_CAUSAL_LM_MODEL_MAP[model_type].from_quantized(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/nas/yuehu/NEW/AutoAWQ/awq/models/base.py", line 498, in from_quantized
model = target_cls.from_config(
^^^^^^^^^^^^^^^^^^^^^^^
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/models/auto/auto_factory.py", line 440, in from_config
return model_class._from_config(config, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 280, in _wrapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2023, in _from_config
model = cls(config, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py", line 4386, in init
self.post_init()
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 1886, in post_init
self.init_weights()
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 3149, in init_weights
self.initialize_weights()
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2502, in initialize_weights
self.smart_apply(self._initialize_weights)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2493, in smart_apply
module.smart_apply(module._initialize_weights)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2493, in smart_apply
module.smart_apply(module._initialize_weights)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2495, in smart_apply
module.smart_apply(fn)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2495, in smart_apply
module.smart_apply(fn)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2495, in smart_apply
module.smart_apply(fn)
[Previous line repeated 1 more time]
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2496, in smart_apply
fn(self)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/modeling_utils.py", line 2470, in _initialize_weights
self._init_weights(module)
File "/root/anaconda3/envs/awq/lib/python3.12/site-packages/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py", line 152, in init_weights
module.weight.data.fill
(1.0)
^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'data'
Fixes # (issue)

this PR can fix the above error.

Who can review?
@ArthurZucker @amyeroberts @qubvel

thanks for your reviewing.

fix the error when loading quantized model by AuotAWQ.
@github-actions

github-actions Bot commented May 8, 2025

Copy link
Copy Markdown
Contributor

Hi 👋, thank you for opening this pull request! The pull request is converted to draft by default. The CI will be paused while the PR is in draft mode. When it is ready for review, please click the Ready for review button (at the bottom of the PR page). This will assign reviewers and trigger CI.

@github-actions
github-actions Bot marked this pull request as draft May 8, 2025 10:04
sync code to modular_qwen2_5_omni.py
@tiger-of-shawn
tiger-of-shawn marked this pull request as ready for review May 8, 2025 10:41
@tiger-of-shawn

Copy link
Copy Markdown
Contributor Author

@MekkCyber here is the new PR to fix CI error.

@MekkCyber MekkCyber left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sounds good ! Pinging @Cyrilvallez since he worked on _init_weights recently

@tiger-of-shawn

Copy link
Copy Markdown
Contributor Author

Could you please review this PR @ArthurZucker @Rocketknight1 , thank you very much.

@ArthurZucker ArthurZucker left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For sure! sorry for the delayed review, thanks 🤗

@ArthurZucker
ArthurZucker merged commit dbc4b91 into huggingface:main May 20, 2025
faaany pushed a commit to faaany/transformers that referenced this pull request May 21, 2025
…ng quantized weights with AutoAWQ. (huggingface#38013)

* Update modular_qwen2_5_omni.py

fix the error when loading quantized model by AuotAWQ.

* Update modeling_qwen2_5_omni.py

sync code to modular_qwen2_5_omni.py
xvyv99 pushed a commit to xvyv99/transformers that referenced this pull request May 21, 2025
…ng quantized weights with AutoAWQ. (huggingface#38013)

* Update modular_qwen2_5_omni.py

fix the error when loading quantized model by AuotAWQ.

* Update modeling_qwen2_5_omni.py

sync code to modular_qwen2_5_omni.py
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants